CSLAP<-read.csv("CSLAP_Dataset_09232019.csv", header=TRUE, stringsAsFactors = F)
CSLAP$Sample_Year<-as.factor(CSLAP$Sample_Year)
CSLAP$Sample_Month<-as.factor(CSLAP$Sample_Month)
#Create TN:TP column
CSLAP$TN_TP<-(CSLAP$TN_mg.L+.01)/(CSLAP$TP_mg.L+.01)
#Read in and merge %ag
Ag<-read.csv("Percent Ag Cover.csv", na.strings=c("", " "))
CSLAP<-merge(CSLAP, Ag, by="Lake_Name", all.x=TRUE)
#Split by `Info_Type`
OWCSLAP<-CSLAP[CSLAP$Info_Type == "OW",]
BSCSLAP<-CSLAP[CSLAP$Info_Type == "BS",]
SBCSLAP<-CSLAP[CSLAP$Info_Type == "SB",]
noSBCSLAP<-CSLAP[CSLAP$Info_Type != "SB",]
redCSLAP<-read.csv("redCSLAP.csv")
names(redCSLAP)[names(redCSLAP) == "Lake_Name.x"] <- "Lake_Name"
names(redCSLAP)[names(redCSLAP) == "Dreissenids.x"] <- "Dreissenids"
redCSLAP$Sample_Year<-as.factor(redCSLAP$Sample_Year)
redCSLAP$Sample_Month<-as.factor(redCSLAP$Sample_Month)
#Create TN:TP column
redCSLAP$TN_TP<-(redCSLAP$TN_mg.L+.01)/(redCSLAP$TP_mg.L+.01)
#Merge %ag
redCSLAP<-merge(redCSLAP, Ag, by="Lake_Name", all.x=TRUE)
#Split by `Info_Type`
redOWCSLAP<-redCSLAP[redCSLAP$Info_Type == "OW",]
redBSCSLAP<-redCSLAP[redCSLAP$Info_Type == "BS",]
redSBCSLAP<-redCSLAP[redCSLAP$Info_Type == "SB",]
rednoSBCSLAP<-redCSLAP[redCSLAP$Info_Type != "SB",]
lmer()TNTP<-lmer(log(TN_TP+.01) ~ Dreissenids + log(CA.SA) + log(Mean_Depth_m) + Percent_Ag + (1|Sample_Year) + (1|Sample_Month) + (1|LakeID) + (1|Sample_Year:LakeID), data=OWCSLAP)
summary(TNTP)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(TN_TP + 0.01) ~ Dreissenids + log(CA.SA) + log(Mean_Depth_m) +
## Percent_Ag + (1 | Sample_Year) + (1 | Sample_Month) + (1 |
## LakeID) + (1 | Sample_Year:LakeID)
## Data: OWCSLAP
##
## REML criterion at convergence: 3172.9
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -8.2481 -0.4424 0.0144 0.4864 8.0077
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.032401 0.18000
## LakeID (Intercept) 0.036200 0.19026
## Sample_Month (Intercept) 0.001351 0.03676
## Sample_Year (Intercept) 0.008238 0.09076
## Residual 0.165055 0.40627
## Number of obs: 2624, groups:
## Sample_Year:LakeID, 344; LakeID, 68; Sample_Month, 6; Sample_Year, 6
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.703403 0.130632 80.476010 20.695 <2e-16 ***
## DreissenidsUninvaded -0.061902 0.082491 94.133562 -0.750 0.4549
## log(CA.SA) 0.044379 0.026535 62.225626 1.673 0.0994 .
## log(Mean_Depth_m) 0.001304 0.040636 65.427991 0.032 0.9745
## Percent_Ag 0.007264 0.002236 67.770001 3.249 0.0018 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU l(CA.S l(M_D_
## DrssndsUnnv -0.660
## log(CA.SA) -0.455 0.043
## lg(Mn_Dpt_) -0.501 0.071 -0.043
## Percent_Ag -0.256 0.314 -0.047 -0.084
plot(TNTP)
qqPlot(resid(TNTP))
## 1395 3666
## 711 1949
glmer()TNTP2<-glmer(TN_TP ~ Dreissenids + log(CA.SA) + log(Mean_Depth_m) + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=OWCSLAP, family=Gamma(link="log"))
## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge with max|grad| = 0.0016124 (tol = 0.001, component 1)
summary(TNTP2)
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: Gamma ( log )
## Formula: TN_TP ~ Dreissenids + log(CA.SA) + log(Mean_Depth_m) + Percent_Ag +
## (1 | LakeID) + (1 | Sample_Year:LakeID)
## Data: OWCSLAP
##
## AIC BIC logLik deviance df.resid
## 17962 18009 -8973 17946 2616
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.8001 -0.4237 -0.0833 0.2894 16.2155
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.08084 0.2843
## LakeID (Intercept) 0.01804 0.1343
## Residual 0.29233 0.5407
## Number of obs: 2624, groups: Sample_Year:LakeID, 344; LakeID, 68
##
## Fixed effects:
## Estimate Std. Error t value Pr(>|z|)
## (Intercept) 2.790367 0.156455 17.835 <2e-16 ***
## DreissenidsUninvaded -0.065651 0.106021 -0.619 0.5358
## log(CA.SA) 0.037277 0.033027 1.129 0.2590
## log(Mean_Depth_m) 0.005015 0.050622 0.099 0.9211
## Percent_Ag 0.005859 0.002786 2.103 0.0355 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU l(CA.S l(M_D_
## DrssndsUnnv -0.702
## log(CA.SA) -0.470 0.033
## lg(Mn_Dpt_) -0.528 0.076 -0.036
## Percent_Ag -0.271 0.316 -0.052 -0.081
## convergence code: 0
## Model failed to converge with max|grad| = 0.0016124 (tol = 0.001, component 1)
plot(TNTP2)
qqPlot(resid(TNTP2))
## 3666 684
## 1949 343
simTNTP <- simulateResiduals(TNTP)
## Warning in checkModel(fittedModel): DHARMa: fittedModel not in class of
## supported models. Absolutely no guarantee that this will work!
## Model family was recognized or set as continuous, but duplicate values were detected in the response. Consider if you are fitting an appropriate model.
plotSimulatedResiduals(simTNTP)
## plotSimulatedResiduals is deprecated, switch your code to using the plot function
-Sample_Year and Sample_Month removed for having low variance
Chl<-lmer(log(Extracted_Chl.a_ug.L+1) ~ Dreissenids + log(CA.SA) + Mean_Depth_m + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=OWCSLAP)
summary(Chl)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula:
## log(Extracted_Chl.a_ug.L + 1) ~ Dreissenids + log(CA.SA) + Mean_Depth_m +
## Percent_Ag + (1 | LakeID) + (1 | Sample_Year:LakeID)
## Data: OWCSLAP
##
## REML criterion at convergence: 3594.6
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.7666 -0.5667 -0.0549 0.4966 5.7200
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.02834 0.1683
## LakeID (Intercept) 0.08726 0.2954
## Residual 0.19763 0.4446
## Number of obs: 2606, groups: Sample_Year:LakeID, 344; LakeID, 68
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 1.280163 0.150389 84.608941 8.512 5.41e-13 ***
## DreissenidsUninvaded 0.046105 0.107739 125.583342 0.428 0.66943
## log(CA.SA) 0.050864 0.039282 58.680926 1.295 0.20045
## Mean_Depth_m -0.034545 0.010296 60.160977 -3.355 0.00138 **
## Percent_Ag 0.003242 0.003174 63.412744 1.021 0.31100
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU l(CA.S Mn_Dp_
## DrssndsUnnv -0.713
## log(CA.SA) -0.522 0.040
## Mean_Dpth_m -0.290 0.008 -0.202
## Percent_Ag -0.325 0.291 -0.055 -0.012
plot(Chl)
qqPlot(resid(Chl))
## 2429 1225
## 1231 612
-Sample_Year and Sample_Month removed for having low variance
Secchi <- lmer(log(Secchi_Depth_m) ~ Dreissenids + log(CA.SA) + Mean_Depth_m + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=OWCSLAP)
summary(Secchi)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(Secchi_Depth_m) ~ Dreissenids + log(CA.SA) + Mean_Depth_m +
## Percent_Ag + (1 | LakeID) + (1 | Sample_Year:LakeID)
## Data: OWCSLAP
##
## REML criterion at convergence: -267.5
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.9556 -0.5342 0.0136 0.5662 4.8113
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.01263 0.1124
## LakeID (Intercept) 0.08874 0.2979
## Residual 0.04109 0.2027
## Number of obs: 2601, groups: Sample_Year:LakeID, 344; LakeID, 68
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 1.390407 0.130971 104.194792 10.616 < 2e-16 ***
## DreissenidsUninvaded -0.111388 0.082310 255.965715 -1.353 0.1772
## log(CA.SA) -0.100121 0.038135 63.183212 -2.625 0.0108 *
## Mean_Depth_m 0.045664 0.009951 63.697069 4.589 2.15e-05 ***
## Percent_Ag -0.001281 0.003028 68.441407 -0.423 0.6735
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU l(CA.S Mn_Dp_
## DrssndsUnnv -0.625
## log(CA.SA) -0.571 0.031
## Mean_Dpth_m -0.320 0.004 -0.200
## Percent_Ag -0.275 0.233 -0.063 -0.017
plot(Secchi)
qqPlot(resid(Secchi))
## 4614 4673
## 2422 2451
-Gamma model with no random effects comes out great, Gamma model with random effects much worse
-Tried log-transforming True Color, but that made residuals and qqPlot worse
-Sample_Year and Sample_Month removed for having low variance
OWTC<-lmer(True_Color_PTU ~ Dreissenids + log(CA.SA) + log(Mean_Depth_m) + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=OWCSLAP)
summary(OWTC)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: True_Color_PTU ~ Dreissenids + log(CA.SA) + log(Mean_Depth_m) +
## Percent_Ag + (1 | LakeID) + (1 | Sample_Year:LakeID)
## Data: OWCSLAP
##
## REML criterion at convergence: 16659.1
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.6963 -0.4677 -0.0608 0.3894 9.8534
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 29.40 5.422
## LakeID (Intercept) 54.34 7.371
## Residual 23.12 4.808
## Number of obs: 2627, groups: Sample_Year:LakeID, 345; LakeID, 68
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 11.214842 4.243693 82.755937 2.643 0.00983 **
## DreissenidsUninvaded 5.835025 2.660055 133.684968 2.194 0.02999 *
## log(CA.SA) 2.457018 0.958047 59.815013 2.565 0.01286 *
## log(Mean_Depth_m) -4.013096 1.456228 61.390628 -2.756 0.00770 **
## Percent_Ag -0.001866 0.079283 64.703351 -0.024 0.98129
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU l(CA.S l(M_D_
## DrssndsUnnv -0.651
## log(CA.SA) -0.501 0.038
## lg(Mn_Dpt_) -0.544 0.059 -0.036
## Percent_Ag -0.231 0.280 -0.056 -0.096
plot(OWTC)
qqPlot(resid(OWTC))
## 3281 4510
## 1695 2395
-update 03/08 boundary (singular) fit Removed random effect of Sample_Year with ~0.00 variance to resolve warning.
SBChl<-lmer(log(ESF_Chl.a_ug.L) ~ Dreissenids + log(CA.SA) + Mean_Depth_m + Percent_Ag + (1|Sample_Month) + (1|LakeID) , data=SBCSLAP)
summary(SBChl)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(ESF_Chl.a_ug.L) ~ Dreissenids + log(CA.SA) + Mean_Depth_m +
## Percent_Ag + (1 | Sample_Month) + (1 | LakeID)
## Data: SBCSLAP
##
## REML criterion at convergence: 992.8
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.92500 -0.60916 0.06149 0.61850 2.22800
##
## Random effects:
## Groups Name Variance Std.Dev.
## LakeID (Intercept) 2.18554 1.4784
## Sample_Month (Intercept) 0.08275 0.2877
## Residual 4.40087 2.0978
## Number of obs: 219, groups: LakeID, 42; Sample_Month, 6
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 5.24989 1.49081 38.13995 3.522 0.00113 **
## DreissenidsUninvaded -0.28234 0.97251 43.86278 -0.290 0.77294
## log(CA.SA) 0.28546 0.33226 34.64026 0.859 0.39617
## Mean_Depth_m -0.12619 0.11900 30.98357 -1.060 0.29717
## Percent_Ag 0.03388 0.02595 35.42128 1.306 0.20007
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU l(CA.S Mn_Dp_
## DrssndsUnnv -0.744
## log(CA.SA) -0.487 0.063
## Mean_Dpth_m -0.485 0.101 0.001
## Percent_Ag -0.462 0.513 -0.074 0.001
plot(SBChl)
qqPlot(resid(SBChl))
## 614 4342
## 44 195
-update 03/08 boundary (singular) fit Removed random effect of Sample_Month with 0.00 variance to resolve warning.
-update 06/05 adding continuous variable of TN:TP as a fixed effect
-removed (LakeID:Sample_Year) because low variance
#Extracting average annual TP for each lake
library(plyr)
avgTNTP<-ddply(OWCSLAP, c("LakeID", "Sample_Year"), summarize,
Mean = mean(TN_TP, na.rm=TRUE))
colnames(avgTNTP)[colnames(avgTNTP)=="Mean"] <- "TN_TP"
#Merge these values to the SBCSLAP df
SBCSLAP<-SBCSLAP[,c(1:44, 46)]
SBCSLAP<-merge(SBCSLAP, avgTNTP, by=c("LakeID", "Sample_Year"), all.x=TRUE, all.y=FALSE)
SBmicro<-lmer(log(ESF_Microcystin_ug.L) ~ Dreissenids + TN_TP + log(CA.SA) + Mean_Depth_m + Percent_Ag + (1|Sample_Year) + (1|LakeID), data=SBCSLAP)
summary(SBmicro)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(ESF_Microcystin_ug.L) ~ Dreissenids + TN_TP + log(CA.SA) +
## Mean_Depth_m + Percent_Ag + (1 | Sample_Year) + (1 | LakeID)
## Data: SBCSLAP
##
## REML criterion at convergence: 175.4
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.30446 -0.54677 0.03348 0.52972 1.59541
##
## Random effects:
## Groups Name Variance Std.Dev.
## LakeID (Intercept) 2.316 1.522
## Sample_Year (Intercept) 5.950 2.439
## Residual 2.290 1.513
## Number of obs: 43, groups: LakeID, 7; Sample_Year, 6
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 17.19767 7.12147 2.01423 2.415 0.136
## DreissenidsUninvaded -1.87235 2.42891 1.02284 -0.771 0.580
## TN_TP -0.05043 0.05720 4.77410 -0.882 0.420
## log(CA.SA) -3.12110 2.31130 1.82983 -1.350 0.320
## Mean_Depth_m -0.77147 0.46167 1.53193 -1.671 0.273
## Percent_Ag -0.06223 0.07037 1.53839 -0.884 0.493
##
## Correlation of Fixed Effects:
## (Intr) DrssnU TN_TP l(CA.S Mn_Dp_
## DrssndsUnnv -0.275
## TN_TP -0.153 0.006
## log(CA.SA) -0.839 -0.177 0.011
## Mean_Dpth_m -0.769 0.154 -0.065 0.593
## Percent_Ag -0.633 0.575 -0.283 0.377 0.462
plot(SBmicro)
qqPlot(resid(SBmicro))
## 60 36
## 23 16
-Sample_Month and Sample_Year removed for low variance
redTNTP<-glmer(TN_TP ~ Dreissenids + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=redOWCSLAP, family=Gamma(link="log"))
summary(redTNTP)
## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: Gamma ( log )
## Formula:
## TN_TP ~ Dreissenids + Percent_Ag + (1 | LakeID) + (1 | Sample_Year:LakeID)
## Data: redOWCSLAP
##
## AIC BIC logLik deviance df.resid
## 4223.4 4250.0 -2105.7 4211.4 612
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.1584 -0.4826 -0.0688 0.3196 11.4263
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.05114 0.2261
## LakeID (Intercept) 0.02977 0.1725
## Residual 0.20316 0.4507
## Number of obs: 618, groups: Sample_Year:LakeID, 80; LakeID, 16
##
## Fixed effects:
## Estimate Std. Error t value Pr(>|z|)
## (Intercept) 2.717645 0.172302 15.773 <2e-16 ***
## DreissenidsUninvaded 0.045465 0.170140 0.267 0.7893
## Percent_Ag 0.015466 0.006712 2.304 0.0212 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU
## DrssndsUnnv -0.730
## Percent_Ag -0.721 0.427
plot(redTNTP)
qqPlot(resid(redTNTP))
## 552 350
## 281 158
-Sample_Year and Sample_Month removed for having low variance
redChl<-lmer(log(Extracted_Chl.a_ug.L+1) ~ Dreissenids + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=redOWCSLAP)
summary(redChl)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(Extracted_Chl.a_ug.L + 1) ~ Dreissenids + Percent_Ag + (1 |
## LakeID) + (1 | Sample_Year:LakeID)
## Data: redOWCSLAP
##
## REML criterion at convergence: 703
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.1216 -0.6131 -0.0398 0.5263 5.7090
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.02612 0.1616
## LakeID (Intercept) 0.02947 0.1717
## Residual 0.15418 0.3927
## Number of obs: 619, groups: Sample_Year:LakeID, 81; LakeID, 16
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 1.209719 0.097822 14.514124 12.367 4.2e-09 ***
## DreissenidsUninvaded -0.035804 0.098511 21.579328 -0.363 0.720
## Percent_Ag -0.002540 0.003759 11.382668 -0.676 0.513
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU
## DrssndsUnnv -0.741
## Percent_Ag -0.727 0.444
plot(redChl)
qqPlot(resid(redChl))
## 391 626
## 175 314
-Sample_Year and Sample_Month removed for having low variance
redSecchi <- lmer(log(Secchi_Depth_m) ~ Dreissenids + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=redOWCSLAP)
summary(redSecchi)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(Secchi_Depth_m) ~ Dreissenids + Percent_Ag + (1 | LakeID) +
## (1 | Sample_Year:LakeID)
## Data: redOWCSLAP
##
## REML criterion at convergence: -48.3
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.7424 -0.5271 -0.0301 0.5525 3.7536
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 0.02067 0.1438
## LakeID (Intercept) 0.09194 0.3032
## Residual 0.04004 0.2001
## Number of obs: 620, groups: Sample_Year:LakeID, 81; LakeID, 16
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 1.441125 0.130482 23.215997 11.045 1.02e-10 ***
## DreissenidsUninvaded -0.137430 0.106535 69.776663 -1.290 0.201
## Percent_Ag -0.002111 0.005635 15.373851 -0.375 0.713
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU
## DrssndsUnnv -0.607
## Percent_Ag -0.693 0.324
plot(redSecchi)
qqPlot(resid(redSecchi))
## 1078 105
## 520 56
-Sample_Year and Sample_Month removed for having low variance
-log-transforming True Color does not improve fit
redOWTC<-lmer(True_Color_PTU ~ Dreissenids + Percent_Ag + (1|LakeID) + (1|Sample_Year:LakeID), data=redOWCSLAP)
summary(redOWTC)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: True_Color_PTU ~ Dreissenids + Percent_Ag + (1 | LakeID) + (1 |
## Sample_Year:LakeID)
## Data: redOWCSLAP
##
## REML criterion at convergence: 3922.9
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.9656 -0.4734 -0.0560 0.4393 4.9069
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year:LakeID (Intercept) 45.19 6.722
## LakeID (Intercept) 13.37 3.656
## Residual 21.71 4.659
## Number of obs: 623, groups: Sample_Year:LakeID, 81; LakeID, 16
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 10.2757 2.4593 15.1672 4.178 0.00079 ***
## DreissenidsUninvaded 6.3058 2.5638 19.3133 2.460 0.02350 *
## Percent_Ag 0.1131 0.0925 13.3744 1.223 0.24247
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU
## DrssndsUnnv -0.760
## Percent_Ag -0.732 0.461
plot(redOWTC)
qqPlot(resid(redOWTC))
## 235 212
## 100 88
-Sample_Year removed for low variance
redSBChl<-lmer(log(ESF_Chl.a_ug.L) ~ Dreissenids + Percent_Ag + (1|Sample_Month) + (1|LakeID) , data=redSBCSLAP)
summary(redSBChl)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(ESF_Chl.a_ug.L) ~ Dreissenids + Percent_Ag + (1 | Sample_Month) +
## (1 | LakeID)
## Data: redSBCSLAP
##
## REML criterion at convergence: 491
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.78450 -0.49874 0.03112 0.62808 2.18538
##
## Random effects:
## Groups Name Variance Std.Dev.
## LakeID (Intercept) 3.3489 1.8300
## Sample_Month (Intercept) 0.2952 0.5434
## Residual 4.2813 2.0691
## Number of obs: 110, groups: LakeID, 9; Sample_Month, 6
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 6.452488 1.826561 13.218613 3.533 0.00359 **
## DreissenidsUninvaded -0.842448 1.364656 22.383307 -0.617 0.54324
## Percent_Ag 0.000857 0.055419 8.252143 0.015 0.98803
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) DrssnU
## DrssndsUnnv -0.729
## Percent_Ag -0.859 0.532
plot(redSBChl)
qqPlot(resid(redSBChl))
## 164 1024
## 18 97
-update 03/08 boundary (singular) fit Removed random effect of Sample_Month with 0.00 variance to resolve warning.
-update 06/05 adding continuous variable of TN:TP as a fixed effect
-removed (LakeID:Sample_Year) because low variance
#Extracting average annual TP for each lake
library(plyr)
redavgTNTP<-ddply(redOWCSLAP, c("LakeID", "Sample_Year"), summarize,
Mean = mean(TN_TP, na.rm=TRUE))
colnames(redavgTNTP)[colnames(redavgTNTP)=="Mean"] <- "TN_TP"
#Merge these values to the SBCSLAP df
redSBCSLAP<-redSBCSLAP[,c(1:52, 54)]
redSBCSLAP<-merge(redSBCSLAP, redavgTNTP, by=c("LakeID", "Sample_Year"), all.x=TRUE, all.y=FALSE)
redSBmicro<-lmer(log(ESF_Microcystin_ug.L) ~ Dreissenids + TN_TP + Percent_Ag + (1|Sample_Year) + (1|LakeID), data=redSBCSLAP)
## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## unable to evaluate scaled gradient
## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge: degenerate Hessian with 1 negative eigenvalues
summary(redSBmicro)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log(ESF_Microcystin_ug.L) ~ Dreissenids + TN_TP + Percent_Ag +
## (1 | Sample_Year) + (1 | LakeID)
## Data: redSBCSLAP
##
## REML criterion at convergence: 144.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.28693 -0.54755 0.04839 0.58461 1.60177
##
## Random effects:
## Groups Name Variance Std.Dev.
## Sample_Year (Intercept) 6.371425 2.52417
## LakeID (Intercept) 0.001968 0.04436
## Residual 2.346597 1.53186
## Number of obs: 35, groups: Sample_Year, 5; LakeID, 3
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 7.402e+00 3.071e+00 6.141e-04 2.410 0.997
## DreissenidsUninvaded -1.127e+00 1.317e+00 8.937e-05 -0.855 1.000
## TN_TP -5.540e-02 6.130e-02 3.539e+00 -0.904 0.423
## Percent_Ag -1.818e-02 6.826e-02 2.738e-04 -0.266 0.999
##
## Correlation of Fixed Effects:
## (Intr) DrssnU TN_TP
## DrssndsUnnv -0.830
## TN_TP -0.261 0.053
## Percent_Ag -0.623 0.714 -0.512
## convergence code: 0
## unable to evaluate scaled gradient
## Model failed to converge: degenerate Hessian with 1 negative eigenvalues
plot(redSBmicro)
qqPlot(resid(redSBmicro))
## 31 22
## 19 16